Subcellular localization of cyclic nucleotide phosphodiesterase type 10A variants, and alteration of the localization by cAMP-dependent protein kinase-dependent phosphorylation.

Our previous studies have suggested that two phosphodiesterase type 10A (PDE10A) variants, PDE10A1 and PDE10A2 transcripts, are mainly expressed in humans and that PDE10A2 and PDE10A3 transcripts are major variants in rats. In the present study, immunoblot analysis demonstrated that PDE10A proteins, especially PDE10A2, are more abundant in membrane fractions than in cytosolic fractions of rat striatum. Recombinant PDE10A1 and PDE10A3 were produced only in cytosolic fractions of transfected PC12h cells. By contrast, recombinant PDE10A2 was present mainly in membrane fractions. This finding agreed well with the result of subcellular fractionation of PDE10A in rat striatum. Immunocytochemical analysis showed that PDE10A2 was localized in the Golgi apparatus of transfected PC12h cells. PDE10A2 was phosphorylated by cAMP-dependent protein kinase (PKA) at Thr16. Interestingly, recombinant protein of wild-type PDE10A2, but not PDE10A2 mutant with an Ala replacement at Thr16, was distributed to cytosolic fractions by co-transfection with a plasmid encoding the catalytic subunit of PKA. A PDE10A2 mutant with Glu substitution at Thr16, which can be a mimic of phosphorylation, was localized in the cytosolic fractions of transfected PC12h cells. These observations implied that phosphorylation of PDE10A2 at Thr16 by PKA caused alteration of subcellular localization of PDE10A2 from the Golgi apparatus to cytosol. It is hypothesized that cAMP signaling in the Golgi area and the cytosol in neurons is controlled through alteration of subcellular localization of PDE10A brought by activation of PKA in response to intracellular elevations of cAMP.

Cyclic nucleotides, cAMP and cGMP, control physiological functions in neuronal networks. Dopamine, adenosine, and vasoactive intestinal peptide are neuronal transmitters and cause elevation of intracellular cAMP levels in striatal neurons (1)(2)(3). Cyclic AMP-dependent protein kinase (PKA), 1 which is activated by cAMP, phosphorylates certain receptors and in-tracellular proteins such as dopamine-and cAMP-related phosphoprotein of M r 32,000 (1) and glutamate receptor (4,5) in striatal neurons. The cellular compartmentalization of cAMP signaling through association with several forms of protein kinase A-anchoring proteins (AKAPs), which interact with the regulatory subunit of PKA, has been discussed (6). Disorder of subcellular localization of PKA and AKAP prevents appropriate differentiation of PC12 cells stimulated by cAMP-producing agents (7). Thus, signal transduction by cyclic nucleotides is regulated by multiple components in neurons.
Cyclic nucleotides are hydrolyzed by phosphodiesterases (PDEs), which consist of 11 families (8,9). We have isolated cDNAs for a dual substrate PDE that hydrolyzes both cAMP and cGMP (PDE10A) from humans and rats (10), and another group has reported mouse PDE10A cDNA (11). K m values of PDE10A for cAMP and cGMP are 0.26 M and 7.2 M, respectively, and V max values for them are slightly different (10). We have previously reported that PDE10A transcripts and their enzymatic activities are abundant in the striatum and testis (12), and more detailed expression patterns of PDE10A transcripts and proteins in rat brain have been recently reported (13). PDE10A is composed of at least 11 alternative splice variants possessing unique amino and carboxyl termini (10,(13)(14)(15). It is suggested that PDE10A1 and PDE10A2 transcripts are major variants in humans and that PDE10A2 and PDE10A3 transcripts are major variants in rats (Fig. 1). The unique amino-terminal region of PDE10A2 is phosphorylated by the catalytic subunit of PKA in vitro (14). In other PDE families, many splice variants show different tissue expression patterns and distinct subcellular localization and are subjected to unique regulation by protein kinases and associated proteins (8,(17)(18)(19)(20). For example, whereas phosphorylation of the amino terminus of the PDE4D3 variant by PKA increases its activity (21)(22)(23), PDE4D3 is inactivated by extracellular signal-regulated kinase-mediated phosphorylation in vivo (24). PDE4A1, PDE4D, and PDE2A are localized in the Golgi apparatus (25)(26)(27)(28)(29)(30)(31), and PDE3B is associated with the endoplasmic reticulum via its amino-terminal hydrophobic domain (32). Thus, it is intriguing that alterations of the cyclic nucleotide-hydrolyzing activity of PDEs by the phosphorylation and compartmentalization of several signaling molecules such as PDEs, AKAPs, and PKA would be involved in the modulation of the subcellular signaling of cyclic nucleotides. However, changes in biological function of PDE10A2 by phosphorylation have not yet been reported. Here, we demonstrate the unique subcellular localization of PDE10A1, PDE10A2, and PDE10A3 variants and propose a novel mechanism for alteration of the subcellular localization of PDE10A. 32 P]ATP were from Amersham Biosciences. Cyclic GMP, cAMP, Crotalus atrox snake venom, erythro-9-(2-hydroxy-3-nonyl)-adenine, rolipram, and calmodulin were purchased from Sigma. Dowex (1 ϫ 8 200 -400) was purchased from Dow Chemical Co.

Materials-[ 3 H]cGMP, [ 3 H]cAMP, and [␥-
Generation of Antisera-Rabbit polyclonal antisera were raised against the poly-L-lysine-based multiple antigen peptide (for PDE10A1) or the keyhole limpet hemocyanin-conjugated peptide (for PDE10A2 and PDE10A3) complexes containing the sequences MRIEERKSQH-LTGL (residues 1-14 of human PDE10A1), MEDGPSNNASCFRRLTE (residues 1-16 of human and rat PDE10A2), and MSNDSPEGAVG-SCNA (residues 1-15 of rat PDE10A3). The peptides were synthesized by a PSSM-8 peptide synthesizer (Shimadzu, Japan) or purchased from Sawady Technology. The complexes were mixed with Freund's complete adjuvant (Sigma) for the first immunization and with Freund's incomplete adjuvant when boosted. After immunizing five times, antisera were collected and stored at Ϫ80°C until use. Purified PDE10A antibody, which recognizes the common region in PDE10A variants, was prepared from the antiserum reported previously (12), using an antigen affinity chromatography from a HiTrap N-hydroxysuccinimide-activated column (Amersham Biosciences), according to the manufacturer's instructions. The antibody was eluted with 0.1 M glycine-HCl (pH 2.7) and then neutralized immediately.
Preparation of PDE Isozymes from Rat Striatum-Male Sprague-Dawley rats obtained from Japan SLC at 10 weeks of age were anesthetized with diethyl ether before striatum was excised and stored at Ϫ80°C until use. A 0.36-g sample of rat striatum was disrupted in 7.2 ml of ice-cold homogenization buffer HB-A (20 mM Tris-HCl, pH 7.5, 2 mM magnesium acetate, 0.3 mM CaCl 2 , 1 mM dithiothreitol, 1.3 mM benzamidine, and 0.2 mM phenylmethylsulfonyl fluoride) by a sonicator (TOMY Seiko, Japan). The homogenates were centrifuged at 100,000 ϫ g for 60 min at 4°C, and the resultant supernatants were designated as cytosolic fractions. The pellets were resuspended with a volume of ice-cold HBT-A buffer (HB-A buffer containing 0.5% Triton X-100) equal to that of the cytosolic fractions. After sonication, the mixture was incubated for 30 min at 4°C and then centrifuged at 100,000 ϫ g for 60 min at 4°C. The resultant supernatants were designated as membrane fractions. The pellets were resuspended with a volume of ice-cold HBT-A buffer equal to that of the cytosolic fractions, designating detergent-insoluble fractions. Lactate dehydrogenase (LDH) activity in each fraction was assayed using the LDH-cytotoxic test (Wako Pure Chemical Industries). Approximately 20% of the total LDH activity was present in membrane fractions in these experiments. The membrane and cytosolic fractions were applied to a HiTrap Q column (Amersham Biosciences) equilibrated in elution buffer A (20 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 1 mM CaCl 2 , and 5 mM benzamidine), with and without 0.5% Triton X-100, respectively. The column was washed with 20 ml of the above buffer used for equilibration, and proteins were then eluted from the column by running a linear NaCl gradient (0 -0.5 M, 100 ml) in the buffer. Fractions (2 ml each) collected on ice were assayed for cGMP and cAMP hydrolytic activities.
PDE Assay-The PDE assay was performed by the radiolabeled nucleotide method. The assay buffer contained 50 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 4 mM 2-mercaptoethanol, 0.33 mg/ml bovine serum albumin (Sigma), enzyme solution, unlabeled cGMP or cAMP, and 12.5 nM [ 3 H]cGMP or 4.88 nM [ 3 H]cAMP. The reaction was started by mixing the substrate into 500 l of the assay buffer, and tubes were incubated at 37°C for 30 min. After boiling for 1.5 min, the mixtures were added to 100 l of 1 mg/ml Crotalus atrox snake venom and incubated at 37°C for 30 min. The reaction was stopped by the addition of 500 l of methanol. The resultant solutions were applied to a Dowex (1 ϫ 8 200 -400) column (5 ϫ 12 mm), and flow-through fractions containing [ 3 H]adenosine or [ 3 H]guanosine were collected. After washing the column with 1.0 ml of methanol, the solutions passing through the column were mixed with the flow-through fractions. Aqueous scintillation mixtures were added to the mixtures, and the radioactivities were then measured.
Chemical compounds were dissolved in Me 2 SO. For studies of the inhibition of PDE activities, inhibitors in Me 2 SO (final concentration 1% (v/v)) were added to the assay buffer containing enzyme and preincubated for 5 min before the reactions were initiated by the addition of substrate.
Expression of PDE10A Variants in PC12h Cells-PC12h cells, a subclone of rat pheochromocytoma PC12 cells, were kindly provided by Drs. Hiroshi Hatanaka and Masashi Yamada of Osaka University (34). PC12h cells (6.3 ϫ 10 5 cells/well) were cultured in poly-L-lysine-coated 6-well plates containing Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum and 5% horse serum at 37°C in 5% CO 2 . 2 g/well of the PDE10A plasmid DNA (pFLAG-h10A1, pFLAG-h10A2, pFLAG-h10A2-T16A, pFLAG-h10A2-T16E, pFLAG-10A2-T16S, pFLAG-10A2-T16D, pFLAG-10A2-T16K, pFLAG-10A2-T16G, pFLAG-h10A⌬13, pFLAG-r10A2, or pFLAG-r10A3) were used for single transfection; and 0.5 g/well of the PDE10A plasmid, 0.5 g/well of the PKA plasmid, pFLAG-cAK-C␣, and 1g/well of the pFLAG-CMV2 vector were used for co-transfection. 1 g/well of the PDE10A plasmid and 1 g/well of the PKA mutant plasmid, pFLAG-cAK-C␣-K169A, was used for co-transfection. All transfections were carried out by the LipofectAMINE 2000 (Invitrogen), according to the manufacturer's instructions. 24 h after transfection, the cells were washed with ice-cold phosphate-buffered saline (PBS) and scraped in 0.3 ml/well of ice-cold HB-A buffer with or without 50 mM NaF, 5 mM ␤-glycerophosphoric acid disodium salt, and 1 mM sodium orthovanadate. In the case of phosphorylation studies on co-transfection of the PDE10A and PKA plasmids, the phosphatase inhibitors described above were added to the HB-A buffer. After cells were disrupted by a sonicator, cytosolic and membrane fractions were prepared by ultracentrifugation using the HB-A buffer without and with 0.5% Triton X-100, respectively, as described above. Less than 10% of the total LDH activity was present in membrane fractions.
Immunoblotting-Cytosolic and membrane fractions prepared from transfected cells or rat striatum were subjected to 7.5% SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Millipore Corp.). After blocking with 4% Block Ace (Snow Brand Milk Products, Tokyo, Japan) overnight, the blots were incubated with antibodies or antiserum and peroxidase-conjugated anti-IgG antibody. Visualization of bands was performed by the ECL Western blotting detection reagents (Amersham Biosciences). To measure relative amounts of the proteins, the autoradiogram was scanned and quantified by a PDI 420oe Scanner (PDI, Huntington, NY) and a Quantity One Program (PDI). Protein concentrations were determined by the Dc protein assay kit (Bio-Rad) using bovine serum albumin.
Immunofluorescence Microscopy-24 h after transfection with expression plasmids, PC12h cells were fixed for 30 min in 4% paraformaldehyde and 4% sucrose in phosphate buffer (pH 7.2) at room temperature. The fixed cells were washed, permeabilized with PBS containing 0.1% Tween 20 for single staining or 0.2% Triton X-100 for double staining, and incubated with anti-FLAG mouse monoclonal antibody (1:500 dilution; Sigma) for 3 h or anti-FLAG rabbit polyclonal antibody (1:160 dilution; Sigma) for 1 h and/or anti-TGN38 (trans-Golgi network 38) monoclonal antibody (1:500 dilution; Transduction Laboratories) for 1 h at room temperature. Primary antibodies were washed with PBS, and then the cells were incubated with FITC-conjugated anti-IgG antibody and/or rhodamine-conjugated anti-IgG antibody (Jackson Laboratories). The cells were washed with PBS and viewed by a confocal laser-scanning microscope (TCS-NT, Leica, Germany). The images shown in the figures are representative of those obtained from at least three independent immunostaining experiments. At least two images per experiment were taken, and more than 10 fields in each experiment were observed.
In Vitro Phosphorylation of PDE10A by PKA-Relative amounts of FLAG-PDE10A proteins produced in COS-7 cells were estimated by immunoblotting using anti-FLAG monoclonal antibody as described previously (14). The amounts of the recombinant proteins were also confirmed by measuring their PDE activities. 48 h after transfection, the cells were washed with ice-cold PBS and scraped in ice-cold homogenization buffer HB-B (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 10 M leupeptin, 10 l/ml aprotinin, and 1% Nonidet P-40). The cells were disrupted by a sonicator, and homogenates were centrifuged at 100,000 ϫ g for 60 min at 4°C. High speed supernatants, which provided equal amounts of wild-type and mutanttype FLAG-tagged human PDE10A2 proteins (FLAG-PDE10A2 and FLAG-PDE10A2-T16A, respectively) were mixed with protein G-Sepharose (Amersham Biosciences) and anti-FLAG monoclonal antibody and incubated at 4°C overnight by rotation. The samples were centrifuged at 5,000 ϫ g for 1 min, and the pellets were washed three times with ice-cold HB-B buffer. The resultant pellets were resuspended with the HB-B buffer and then added to each tube. After centrifugation, the immunoprecipitated proteins were subjected to immunoblotting and the PDE assay. FLAG-tagged PDE10A proteins providing equal optical densities by immunoblotting exhibited similar enzymatic activities (data not shown). The immunoprecipitated proteins showing equal amounts were phosphorylated at 30°C for various times in reaction buffer (20 mM Tris-HCl, pH 7.5, 20 mM magnesium acetate, 200 M ATP, 6.7 nM [␥-32 P]ATP), 5 mM ␤-glycerophosphoric acid disodium salt, and 1 mM sodium orthovanadate) containing various concentrations of the catalytic subunit of bovine heart PKA (Promega). The phosphorylated mixtures were centrifuged at 5,000 ϫ g for 1 min. The pellets were resuspended in the Laemmli buffer for SDS-PAGE (35) and boiled. The samples were centrifuged at 5,000 ϫ g for 1 min, and the supernatants were subjected to SDS-PAGE and autoradiography.
Isolation of Phosphorylated PDE10A-First, the specificity of isolation of phosphorylated PDE10A2 with a PhosphoProtein TM affinity column (Qiagen) was verified. FLAG-tagged PDE10A2 proteins produced in transfected COS-7 cells were immunoprecipitated with or without FLAG-tagged PKA catalytic subunit showing equal amounts of proteins by an anti-FLAG monoclonal antibody and protein G-Sepharose in homogenization buffer HB-C (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, Complete TM (Roche Molecular Biochemicals), and 0.1% CHAPS) as described above. The immunoprecipitated proteins were phosphorylated at 30°C for 1 h in reaction buffer (20 mM Tris-HCl, pH 7.5, 20 mM magnesium acetate, 200 M ATP, 6.7 nM [␥-32 P]ATP, and phosphatase inhibitor mix (Sigma)). After washes with the lysis buffer of a Phospho-Protein TM purification kit, FLAG-tagged proteins were eluted with 200 g/ml FLAG peptide in the lysis buffer containing 0.25% CHAPS. The phosphorylated proteins in eluted solutions were separated with a PhosphoProtein TM affinity column according to the manufacturer's instructions, and eluted and flow-through fractions were concentrated to 400 l using an Amicon filtration cell equipped with a PM-30 membrane (Amicon). These concentrated fractions were analyzed by immunoblotting and autoradiography. The autoradiography was performed by BAS2000 (Fujifilm).
To isolate phosphorylated proteins from transfected PC12h cells, affinity column chromatography was performed using a PhosphoProtein TM purification kit. Transfected cells (3.8 ϫ 10 6 cells) were disrupted by a sonicator in 0.9 ml of the lysis buffer of the kit without detergent, and homogenates were centrifuged at 100,000 ϫ g for 60 min at 4°C. The supernatants were used as cytosolic fractions. The pellets were resuspended with 0.9 ml of the lysis buffer containing 0.25% CHAPS. After sonication, the mixture was incubated for 30 min at 4°C and then centrifuged at 100,000 ϫ g for 60 min at 4°C. The resultant supernatants were designated as membrane fractions. The pellets were resuspended with 0.9 ml of the lysis buffer containing 0.25% CHAPS, resulting in insoluble fractions. Each fraction was subjected to immunoblotting using an anti-FLAG antibody to confirm subcellular localization of the PDE10A proteins. The cytosolic and membrane fractions were mixed, and the protein concentrations of the mixtures were adjusted to 0.1 mg/ml by adding the lysis buffer. The solutions were loaded onto the PhosphoProtein TM purification column of the kit and washed with the lysis buffer, and then phosphorylated proteins were eluted with the elution buffer of the kit containing free phosphate. Eluted and flow-through fractions were concentrated to 500 l using an Amicon filtration cell equipped with a PM-30 membrane (Amicon), and then immunoblotting of these fractions was performed.

Distribution of PDE10A in Rat Striatum-
The presence of PDE10A in high speed supernatants (cytosol), detergent-soluble fractions (membrane), and detergent-insoluble fractions of rat striatum were investigated by immunoblot analysis. Using affinity chromatography, we first isolated anti-PDE10A antibody recognizing several PDE10A variants from the anti-PDE10A antiserum produced previously (12). The proteins in cytosolic, membrane, and insoluble fractions were subjected to immunoblot analysis with purified antibody, according to their original protein contents in each fraction. The anti-PDE10A antibody showed strong staining of an ϳ87-kDa band in the proteins from membrane fractions ( Fig. 2A). A moderate signal of the same protein size was observed in both cytosolic and insoluble fractions. Thus, PDE10A protein was particularly abundant in membrane fractions. No signal was observed in the fractions by purified anti-PDE10A antibody preadsorbed with antigen peptide. LDH activity was predominantly detected in cytosolic fractions as shown in Fig. 2A.
Characterization of PDE10A Variants and PDE Isozymes in Rat Striatum-HiTrap Q column chromatography was carried out to identify PDE isozymes in both cytosolic and membrane fractions of rat striatum under conditions similar to those reported previously (12). Fig. 2, B and C, shows the elution profiles of cytosolic and membrane fractions, respectively. Proteins immunoreactive with purified anti-PDE10A antibody were observed only in fractions 16 -20 (cytosolic fractions) and fractions 14 -20 (membrane fractions), indicating that cAMP hydrolytic activities in these fractions contained PDE10A (Fig.  2, B and C). Moreover, immunoreactive signals in membrane fractions were much stronger than those in cytosolic fractions. These results indicated that PDE10A protein was highly abundant in membrane fractions of rat striatum compared with that in cytosolic fractions. With regard to other PDE isozymes in cytosolic fractions, cAMP hydrolytic activities (fractions 28 -42) eluting at ϳ0.3 M NaCl were inhibited by a PDE4 inhibitor, rolipram (10 M), but not by 1 M cGMP, demonstrating that these fractions contained PDE4. The cGMP hydrolytic activities in fractions 32-46 (cytosolic fractions) were increased ϳ3-20-fold by the addition of Ca 2ϩ and calmodulin, indicating that these fractions contained Ca 2ϩ /calmodulin-dependent PDE (PDE1). With regard to membrane fractions of rat striatum, cAMP hydrolytic activities in fractions 23-30 were activated slightly by 1 M cGMP and inhibited by a PDE2 inhibitor, erythro-9-(2-hydroxy-3-nonyl)-adenine (10 M), indicating the presence of PDE2 in these fractions. The cGMP hydrolytic activities in fractions 29 -35 of membrane fractions were stimulated slightly by Ca 2ϩ and calmodulin, suggesting that these fractions contained PDE1.
Detection of PDE10A Variants in Rat Striatum-Polyclonal antisera toward the synthetic peptides corresponding to the unique amino-terminal sequences of human PDE10A1, human and rat PDE10A2, and rat PDE10A3 were produced. The antigenic specificity was tested using PDE10A variants produced in COS-7 cells. The cAMP-PDE activities of the cells expressing each PDE10A variant were ϳ70-fold higher than those of mock-transfected cells. Production of the PDE10A proteins seemed to be almost the same among the transfected cells carrying each of the three PDE10A constructs. Extracts of transfected COS-7 cells showing equal levels of PDE10A activities were electrophoretically separated, blotted onto membranes, and then incubated with a purified anti-PDE10A antibody and each PDE10A variant-specific antiserum. Results using purified anti-PDE10A antibody showed that all PDE10A variants gave equal optical densities (Fig. 3A). Each antiserum specifically detected a PDE10A variant carrying an antigen peptide but did not detect other PDE10A variants. The equal optical intensities were obtained after 15-s exposure with anti-PDE10A antibody and anti-PDE10A1 antiserum and 5-s exposure with anti-PDE10A2 and anti-PDE10A3 antisera in immunoblot analysis, indicating that titers of antisera against PDE10A2 and PDE10A3 were about 3-fold higher than those of the anti-PDE10A antibody and the anti-PDE10A1 antiserum.
To determine which type of PDE10A variant is present in rat striatum, we performed immunoblotting using the above antibody and antisera (Fig. 3B). The antibody prepared against a common region of all PDE10A variants showed immunoreactivity mainly in membrane fractions. Interestingly, immunoblotting using anti-PDE10A2 serum revealed that PDE10A2 was predominantly expressed in membrane fractions but not in cytosolic fractions. No signal was observed in the fractions using either a purified anti-PDE10A antibody or anti-PDE10A2 antiserum preadsorbed with the antigen peptides. By contrast, PDE10A1 and PDE10A3 proteins were not detected in rat striatum using anti-PDE10A1 and anti-PDE10A3 antisera under these conditions.
Distribution of Recombinant PDE10A Variants in Mammalian Cells-PC12 and PC12h cells are preferred to investigate signal transduction in nigra and striatal neurons, because these cell lines respond to several neurotransmitters including dopamine and adenosine (36,37). We examined the possible presence of the three variants of PDE10A transcripts in untransfected PC12h using PCR analysis. PDE10A2 and PDE10A3 transcripts, but not PDE10A1 transcripts, were observed in this cell line, whereas the activity of PDE10A separated by chromatography was too low to be detected (data not shown). In order to examine subcellular distribution of PDE10A variants, expression plasmids for recombinant FLAGtagged PDE10A variants (human PDE10A1, human PDE10A2, human PDE10A⌬13, rat PDE10A2, and rat PDE10A3; see Fig.  1) were transfected into PC12h cells, and the cytosolic and membrane fractions of the cells were separated by ultracentrifugation using buffers lacking or containing detergent (0.5% Triton X-100), respectively. As shown in Fig. 4A, recombinant human PDE10A1 and rat PDE10A3 expressed in transfected PC12h cells were present only in cytosolic fractions. Human and rat PDE10A2 proteins were primarily localized in membrane fractions. Human and rat PDE10A2 proteins contain the same amino-terminal sequence, which is distinct from those of PDE10A1 and PDE10A3. To study the role of the amino-terminal sequences of PDE10A variants in the subcellular distribution of these proteins, we produced a mutant of human PDE10A, termed PDE10A⌬13 (Fig. 1), lacking the unique amino-terminal sequence. It is intriguing that PDE10A⌬13 expressed in transfected PC12h cells was detected in cytosolic fractions, suggesting that the unique amino-terminal portion of PDE10A2 serves as an anchor necessary for association with the membrane.
The membrane fractions from PDE10A2-expressing PC12h cells and rat striatum were treated with varying concentrations of NaCl to determine whether or not PDE10A2 could be released from the membrane (Fig. 4B). PDE10A2 prepared from neither PC12h cells nor striatum was released from the membrane by NaCl, indicating that PDE10A2 does not interact with the membrane as a peripheral protein through ionic interactions.
Intracellular Localization of PDE10A Variants in Transfected PC12h Cells-To further investigate intracellular localization of three types of FLAG-tagged PDE10A variants expressed in transiently transfected PC12h cells, immunofluorescent staining analyses were performed on fixed and permeabilized cells with anti-FLAG and FITC-labeled anti-IgG antibodies. Fig. 5, A-C, shows that PDE10A1, PDE10A3, and PDE10A⌬13 were localized in the cytosol of the transfected cells. By contrast, in PDE10A2-expressing cells, staining was apparent in the perinuclear region (Fig. 5D). The perinuclear signal of PDE10A2 overlapped with that of TGN38, which is known to be localized in the trans-Golgi network (Fig. 5, E-G). These findings indicated intracellular localization of PDE10A2 to the Golgi apparatus of PC12h cells. No specific immunofluorescence was found in untransfected cells (Fig. 5H).
Alteration of Subcellular Localization of PDE10A Variants by Active and Inactive PKA-Our previous report demonstrated that PDE10A2 has a motif for PKA phosphorylation (Arg-Arg-Leu-Thr 16 ) in its amino terminus and that PDE10A2, but not another PDE10A variant having a different amino terminus, is phosphorylated by PKA (14). We thus examined the effects of PKA-mediated phosphorylation on subcellular localization of recombinant PDE10A variants. First, to confirm the PKA-phosphorylation site of PDE10A2, an in vitro kinase assay was performed using the wild-type human PDE10A2 (PDE10A2-WT) and its mutant with an Ala substitution at Thr 16 (PDE10A2-T16A). PDE10A2-WT was phosphorylated by the catalytic subunit of PKA in a concentration-dependent manner (Fig. 6A). PKA-mediated 32 P incorporation of PDE10A2-T16A was significantly reduced compared with that of PDE10A2-WT using various concentrations of PKA. Moreover, phosphorylation of PDE10A2-WT was progressively increased during a 30-min incubation, and the phosphorylation level of PDE10A2-T16A was still low even at 30 min of incubation (Fig. 6B). These observations indicated that Thr 16 of PDE10A2 was a significant target for phosphorylation by PKA in vitro.
The expression plasmid coding for PKA catalytic subunit was co-transfected with the PDE10A plasmids, and subcellular localization of PDE10A in cytosolic and membrane fractions was analyzed by immunoblotting. Co-expression of a FLAG-tagged The pellets were resuspended with 0.6 ml of HB-A buffer containing 0.5% Triton X-100. After sonication, the mixture was incubated for 30 min at 4°C and then ultracentrifuged. The secondary supernatants were stored as supernatant 2 (Sup. 2). The same volumes of the supernatants 1 and 2 were loaded onto SDS-PAGE, and immunoblotting was performed with anti-FLAG antibody for recombinant PDE10A2 or anti-PDE10A2 antiserum for rat striatal PDE10A2.

FIG. 5. Intracellular localization of FLAG-tagged PDE10A variants in transiently transfected PC12h cells.
Immunocytochemical detection of FLAG-tagged proteins of human PDE10A1 (A), rat PDE10A3 (B), human PDE10A⌬13 (C), and human PDE10A2 (D) in transfected PC12h cells was performed using anti-FLAG mouse monoclonal antibody and FITC-labeled anti-mouse IgG antibody as described under "Experimental Procedures." PC12h cells expressing PDE10A2 were double-stained with anti-FLAG rabbit polyclonal and anti-TGN38 mouse antibodies and then visualized with anti-FITC-conjugated rabbit IgG (E) and anti-rhodamine-conjugated mouse IgG (F) antibodies, respectively. G, a merged image of E and F. No specific immunofluorescence was observed in untransfected PC12h cells as shown in H. All images were taken with a confocal microscope. catalytic subunit of PKA with recombinant PDE10A variants in PC12h cells changed subcellular localization of human and rat PDE10A2 from membranes to cytosol, whereas PDE10A2 was present in membrane fractions in the absence of PKA (Fig.  6C). By contrast, subcellular localization of PDE10A1 and PDE10A3 lacking a PKA phosphorylation motif was not affected by co-expression of PKA. PDE10A2-T16A was also localized to membranes of transfected PC12h cells. Interestingly, subcellular localization of PDE10A2-T16A was not changed by co-expression of PKA in cells, suggesting that PKA-mediated phosphorylation of Thr 16 in the unique amino terminus of PDE10A2 alters subcellular localization of the protein.
The effect of inactive PKA catalytic subunit having Ala substitution at Lys 168 in the ATP-binding site on subcellular localization of recombinant PDE10A2 protein in PC12h cells was examined. In contrast to the case of the active PKA catalytic subunit, the inactive PKA catalytic subunit did not alter the membrane localization of PDE10A2 in cells (Fig. 6D).
Phosphorylation of PDE10A2 by PKA Co-expression in Transfected PC12h Cells-To determine whether PDE10A2 is phosphorylated by co-expression of the PKA catalytic subunit in transfected PC12h cells, a PhosphoProtein TM purification column was used to separate phosphorylated proteins. First, the specificity of PhosphoProtein TM purification procedures for phosphorylated PDE10A2 was verified using FLAG-tagged PDE10A2 phosphorylated by PKA catalytic subunit in vitro. Whereas nonphosphorylated PDE10A2 was found in flowthrough fractions, phosphorylated PDE10A2 that was labeled with 32 P was eluted in phosphorylated protein fractions (Fig.  7A), indicating that the affinity column is applicable for separation of phosphorylated PDE10A2.
In vivo phosphorylation of PDE10A2 was confirmed as follows. After transfection into PC12h cells, cytosolic, membrane, and insoluble fractions were prepared with 0.25% CHAPS buffer as described under "Experimental Procedures" and subjected to immunoblotting. Fig. 7B shows that PDE10A2 was predominantly localized in cytosol and membrane fractions in the presence and absence of PKA catalytic subunit, respectively, under this condition. A 0.25% CHAPS buffer provided the same results as a buffer containing 0.5% Triton X-100, as described in the legends to Figs. 2A and 4A. After cytosolic fractions were mixed with membrane fractions, the mixture was loaded onto the PhosphoProtein TM purification column. PDE10A2 proteins in flow-through fractions and eluates were analyzed by immunoblotting. As shown in Fig. 7C, PDE10A2 proteins produced without PKA co-expression appeared mainly in flow-through fractions, which contained nonphosphorylated proteins, whereas PDE10A2 co-expressed with PKA catalytic subunit was eluted from the column with elution buffer. These rat PDE10A2 (r10A2), and rat PDE10A3 (r10A3) in combination with (ϩ) or without (Ϫ) an expression plasmid for a FLAG-tagged catalytic subunit of PKA (see "Experimental Procedures"). 24 h after transfection, cytosolic and membrane fractions were prepared in equal volumes of the solution, subjected to SDS-PAGE, and analyzed by immunoblotting using an anti-FLAG antibody.  findings strongly implied that cytosolic PDE10A2 produced in the presence of PKA catalytic subunit was phosphorylated, whereas PDE10A2 localized in membranes without PKA coexpression was not phosphorylated.
Localization of PDE10A2 Mutants at Thr 16 in Transiently Transfected PC12h-In order to confirm the effect of phosphorylation of PDE10A2 at Thr 16 on subcellular localization, subcellular localization of a mutant PDE10A2 (T16E) with Glu replacing Thr 16 , which mimics phosphorylation of this site, was examined. The mutant protein was identified in cytosolic fractions of transfected PC12h cells (Fig. 8), whereas wild type PDE10A2 was expressed in membrane fractions. Replacement of Thr 16 of PDE10A2 with Ser (T16S), Asp (T16D), Lys (T16K), or Gly (T16G) did not affect membrane localization of the PDE10A2. These findings suggested that PDE10A2 is localized to the cytosolic fractions by PKA-mediated phosphorylation of Thr 16 .

Immunocytochemical Analysis of Intracellular Localization of PDE10A2 Co-expressed with PKA in Transfected PC12h
Cells-Intracellular localization of FLAG-tagged wild-type and mutant-type PDE10A2 proteins in transfected PC12h cells was investigated using immunofluorescent staining analysis. Wildtype PDE10A2 and a PDE10A2 mutant with Ala substitution at Thr 16 were localized in the Golgi apparatus of transfected cells (Fig. 9, A-F). The Golgi signal of PDE10A2 overlapped with that of TGN38 (Fig. 9, C and F). Wild-type PDE10A2 co-expressed with histidine-tagged PKA catalytic subunit and a PDE10A2 mutant with a Glu substitution at Thr 16 exhibited cytosolic localization of PDE10A2 (Fig. 9, G and H). Immunoblotting showed that wild type PDE10A2 co-expressed with histidine-tagged PKA catalytic subunit was present in cytosolic fractions (data not shown). These results were in accordance with those obtained by immunoblotting. DISCUSSION We demonstrated here that PDE10A protein is abundant in membrane fractions compared with that in the cytosolic fractions of rat striatum. Moreover, PDE10A was revealed to be a major cAMP-hydrolyzing PDE in membrane fractions of rat striatum under these conditions. On the other hand, levels of PDE10A activities were almost equal to those of PDE4 in cytosolic fractions. Regarding cGMP-hydrolyzing activities, Ca 2ϩ -and calmodulin-activated PDE (PDE1) was rich in rat striatum. PDE10A2 and PDE10A3 transcripts are major variants in rat brain (12). The presence of a PDE10A2 protein exclusively in membrane fractions of rat striatum indicated that this variant is a membrane-associated PDE. By contrast, an immunoreactive protein of PDE10A3 was detected in neither cytosolic nor membrane fractions using the anti-PDE10A3 antibody, although PDE10A3 transcripts were detected in rat striatum by PCR (data not shown). This inconsistency concerning the existence of PDE10A3 transcripts but lack of enzyme protein in the striatum may be due to exclusion by translational regulation, protein degradation of PDE10A3, or translation of PDE10A3 from another initiation site downstream of the putative first methionine of PDE10A3.
Three major PDE10A variants, PDE10A1, PDE10A2, and PDE10A3, possess unique amino termini. Many splice variants of PDEs, other than PDE10A, having unique amino and carboxyl termini show distinct subcellular localization (17,23,25,39). We demonstrated here that PDE10A2 was predominantly present in membrane fractions of transiently transfected PC12h cells. This observation was in accordance with results obtained by chromatography of extracts of striatum. By con- FIG. 7. Separation of phosphorylated PDE10A2 by Phospho-Protein TM affinity column. A, FLAG-tagged PDE10A2 from transfected COS-7 cells was immunoprecipitated with (ϩ) or without (Ϫ) FLAG-tagged PKA catalytic subunit using an anti-FLAG antibody and protein G-Sepharose, and the resultant immunoprecipitants were subjected to an in vitro kinase assay (see "Experimental Procedures"). After elution of recombinant proteins with FLAG peptide from the Sepharose, phosphorylated proteins were separated using a PhosphoProtein TM purification kit (Qiagen). PDE10A2 proteins in flow-through (FT) and eluted (E) fractions were detected by immunoblotting (IB) with an anti-FLAG antibody (upper panel) and autoradiography (lower panel). B, PC12h cells were co-transfected with an expression plasmid for FLAG-tagged PDE10A2 in combination with (ϩ) or without (Ϫ) an expression plasmid for a FLAG-tagged catalytic subunit of PKA. 24 h after transfection, cytosolic (C), membrane (M), and insoluble (I) fractions were prepared and analyzed by immunoblotting using an anti-FLAG antibody. C, phosphorylated and nonphosphorylated proteins in extracts of the cells were separated using a PhosphoProtein TM Purification Kit (see "Experimental Procedures"). Proteins in flow-through (FT) and eluted (E) fractions were analyzed by immunoblotting using an anti-FLAG antibody. The experiments were performed three times with independent transfections.
FIG. 8. Analyses of subcellular distribution of FLAG-tagged PDE10A2 mutants at Thr 16 . PC12h cells were transfected with expression plasmids for FLAG-tagged PDE10A2 proteins: pFLAG-h10A2 (T16; wild type), pFLAG-h10A2-T16S (T16S; Ser substitution at Thr 16 ), pFLAG-h10A2-T16E (T16E; Glu substitution at Thr 16 ), pFLAG-h10A2-T16D (T16D; Asp substitution at Thr 16 ), pFLAG-h10A2-T16K (T16K; Lys substitution at Thr 16 ), and pFLAG-h10A2-T16G (T16G; Gly substitution at Thr 16 ). 24 h after transfection, cytosolic and membrane fractions were prepared by centrifugation, subjected to SDS-PAGE, and analyzed by immunoblotting using an anti-FLAG antibody. A representative experiment of the three performed is reported. Cytosolic and membrane fractions are represented by C and M, respectively. trast, PDE10A1 and PDE10A3 appeared only in cytosol of transfected PC12h cells. PDE10A1 and PDE10A3, which were expected to be species-specific and major splice variants found in humans and rats, respectively (10,12,14), might play the same physiological role in degrading cytosolic cAMP in each species. The fluorescent immunocytochemistry of transiently transfected PC12h cells demonstrated that localization of PDE10A2 was restricted to the Golgi apparatus. Myomegalin, a soluble fragment of AKAP450, and muscle-selective AKAP were identified as anchoring proteins of PDE4D to the Golgi/ centrosomal region (30,31,40). More recently, tryptophananchoring phosphatidic acid-selective binding domain 1 of PDE4A1 was demonstrated to be associated with phosphatidic acid of plasma membrane (39). PDE10A2 has no transmembrane domain in its unique amino-terminal region. Use of salt washes of membranes demonstrated that membrane PDE10A2 is not a peripheral protein bound through ionic interactions. Therefore, PDE10A2 may be associated with anchoring proteins including AKAP, or it may be anchored by other mechanisms such as association to the membrane through a carbohydrate chain at a putative N-glycosylation site (Asn 8 ) in the amino-terminal region of PDE10A2.
Some studies of translocation and alteration of localization of PDEs have been reported. After prolonged thyroid-stimulating hormone stimulation, PDE4D present in the perinuclear region of FRTL-5 thyroid cells diffuses throughout the cytoplasm (25). In rat aortic vascular smooth muscle cells, PDE4D3 is translocated from particulate to cytosolic fractions by co-stimulation with forskolin and phorbol 12-myristate 13-acetate (23). It is plausible that phosphorylation of PDE10A2 by PKA in PC12h cells causes alteration of PDE10A2 localization to the cytosol. Subcellular localization of PDE10A2-T16A lacking a PKA phosphorylation site was not changed to cytosol in the presence of the catalytic subunit of PKA, and as expected, PDE10A2-T16E, which contains a negatively charged amino acid residue at Thr 16 , showed cytosolic localization in transfected PC12h cells.
However, another PDE10A2 mutant (PDE10A2-T16D) having a negatively charged amino acid, aspartic acid, replacing Thr 16 failed to change the subcellular localization of the protein. From these observations, glutamic acid is likely to be structurally more suitable for mimicking phosphorylated threonine of PDE10A2 than is aspartic acid. In addition, cytosolic PDE10A2 co-expressed with the catalytic subunit of PKA in PC12h cells was eluted in phosphoprotein fractions using a PhosphoProtein TM affinity column. Thus, phosphorylation of Thr 16 in the amino-terminal region of PDE10A2 is required for the alteration of subcellular localization.
The physiological meaning of compartmentalization of PDEs in the Golgi area of neurons is not well understood. One role of PDE10A2 in the perinuclear region of cells is coordination of cyclic nucleotide signaling in this local space. The RII subunit of PKA binds to Neurobeachin, which anchors PKA to the Golgi/centrosome of PC12 cells and neuronal cells (41). The ␣ isoform catalytic subunit of PKA translocates from the Golgi area to the nucleus in neuronal cells via intracellular cAMP elevation induced by 24-h stimulation by ethanol, and phosphorylates the cAMP response element binding factor in the nucleus (42,43). There are some reports concerning cAMP signaling via PKA activation in the Golgi apparatus of neurons. The presence of PDE10A2 in the Golgi area of transfected PC12h cells and alteration of subcellular localization of PDE10A2 to cytosol in response to its phosphorylation by PKA might modulate PKA activation in the Golgi area of neuronal cells. Another possibility is that the Golgi apparatus functions as a reservoir for PDE10A2 in the resting state of cells and that PDE10A2 is recruited from the Golgi apparatus to the cytosol in response to cytosolic cAMP elevation and PKA activation, in order to reduce cAMP signaling in the cytosol. In striatal neurons, several neurotransmitters such as dopamine elevate cytosolic cAMP levels via their specific receptors, and PKA activated through cytosolic cAMP elevation phosphorylates dopamine-and cAMP-related phosphoprotein of M r 32,000 in the striatum (3,4). Since PDE10A activity in membrane fractions of rat striatum is much higher than cytosolic cAMP-PDE activities provided by PDE10A and PDE4, etc., the change in PDE10A2 localization from the Golgi apparatus to the cytosol would be expected to dramatically increase the rate of hydrolysis of cytosolic cAMP. This change in PDE10A2 distribution by PKA-mediated phosphorylation would represent negative feedback regulation of cAMP signaling by reduction of cytosolic cAMP. In conclusion, this study demonstrated that subcellular localization of PDE10A2 to the Golgi apparatus is altered by phosphorylation of this enzyme by PKA. Alteration of subcellular localization of PDE10A2 in the presence and absence of activated PKA supports the thesis that compartmentalized cAMP signaling in the brain can be systematically regulated by several factors, including PDE10A.